Local and systemic N signaling are involved in Medicago truncatula preference for the most efficient Sinorhizobium symbiotic partners

Authors

  • Gisèle Laguerre,

    1. INRA, USC 1242, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    2. IRD, UMR 113, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    3. CIRAD, UMR 113, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    4. SupAgro, UMR 113, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    5. UM2, UMR 113, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    Search for more papers by this author
  • Karine Heulin-Gotty,

    1. INRA, USC 1242, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    2. IRD, UMR 113, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    3. CIRAD, UMR 113, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    4. SupAgro, UMR 113, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    5. UM2, UMR 113, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    Search for more papers by this author
  • Brigitte Brunel,

    1. INRA, USC 1242, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    2. IRD, UMR 113, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    3. CIRAD, UMR 113, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    4. SupAgro, UMR 113, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    5. UM2, UMR 113, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    Search for more papers by this author
  • Agnieszka Klonowska,

    1. INRA, USC 1242, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    2. IRD, UMR 113, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    3. CIRAD, UMR 113, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    4. SupAgro, UMR 113, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    5. UM2, UMR 113, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    Search for more papers by this author
  • Antoine Le Quéré,

    1. INRA, USC 1242, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    2. IRD, UMR 113, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    3. CIRAD, UMR 113, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    4. SupAgro, UMR 113, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    5. UM2, UMR 113, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    Search for more papers by this author
  • Pascal Tillard,

    1. INRA, UMR 5004, Biochimie et Physiologie Moléculaire des Plantes, F-34000 Montpellier, France
    2. CNRS, Biochimie et Physiologie Moléculaire des Plantes, F-34000 Montpellier, France
    3. SupAgro, Biochimie et Physiologie Moléculaire des Plantes, F-34000 Montpellier, France
    4. UM2, Biochimie et Physiologie Moléculaire des Plantes, F-34000 Montpellier, France
    Search for more papers by this author
  • Yves Prin,

    1. INRA, USC 1242, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    2. IRD, UMR 113, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    3. CIRAD, UMR 113, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    4. SupAgro, UMR 113, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    5. UM2, UMR 113, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    Search for more papers by this author
  • Jean-Claude Cleyet-Marel,

    1. INRA, USC 1242, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    2. IRD, UMR 113, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    3. CIRAD, UMR 113, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    4. SupAgro, UMR 113, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    5. UM2, UMR 113, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    Search for more papers by this author
  • Marc Lepetit

    1. INRA, USC 1242, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    2. IRD, UMR 113, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    3. CIRAD, UMR 113, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    4. SupAgro, UMR 113, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    5. UM2, UMR 113, Symbioses Tropicales et Méditerranéennes, F-34000 Montpellier, France
    Search for more papers by this author

Author for correspondence:
Gisèle Laguerre
Tel: +33 4 67 59 38 62
Email: gisele.laguerre@supagro.inra.fr

Summary

  • Responses of the Medicago truncatulaSinorhizobium interaction to variation in N2-fixation of the bacterial partner were investigated.
  • Split-root systems were used to discriminate between local responses, at the site of interaction with bacteria, and systemic responses related to the whole plant N status.
  • The lack of N acquisition by a half-root system nodulated with a nonfixing rhizobium triggers a compensatory response enabling the other half-root system nodulated with N2-fixing partners to compensate the local N limitation. This response is mediated by a stimulation of nodule development (number and size) and involves a systemic signaling mechanism related to the plant N demand. In roots co-infected with poorly and highly efficient strains, partner choice for nodule formation was not modulated by the plant N status. However, the plant N demand induced preferential expansion of nodules formed with the most efficient partners when the symbiotic organs were functional. The response of nodule expansion was associated with the stimulation of symbiotic plant cell multiplication and of bacteroid differentiation.
  • A general model where local and systemic N signaling mechanisms modulate interactions between Medicago truncatula and its Sinorhizobium partners is proposed.

Introduction

Leguminous plants display the capacity to acquire gaseous nitrogen (N2) through the establishment of a symbiotic association with soil bacteria called rhizobia which proliferate and differentiate inside root nodules. The mutual benefit of this association is based on nutritional exchanges between plants and bacteria: atmospheric N2 is fixed by rhizobia and released as NH4+ to the plant, whereas the host plant provides C metabolites to the bacteria. However, in soil, the roots are often subjected to stresses that temporally and locally limit N acquisition, especially N2 fixation which is highly sensitive to environmental factors. In addition, legumes are usually nodulated by indigenous rhizobial populations displaying genetic polymorphism associated with variability in N2 fixation effectiveness. Plant growth will therefore depend on its ability to offset this N deficit by increasing nitrogen acquisition capacity in unstressed root parts.

Regulations responsible for the adjustment of symbiotic N acquisition capacities to the N demand of the whole plant have been characterized in the model legume Medicago truncatula (Mt) using split-root systems (Ruffel et al., 2008; Jeudy et al., 2010). When nodulated plants are supplied by mineral N, a systemic N ‘satiety’ mechanism represses specific N2 fixation activity (SNA) of the symbiotic organs, adjusting their N acquisition capacities to the N demand of the whole plant (Ruffel et al., 2008). By contrast, plant N limitation does not trigger upregulation of nodule SNA and therefore results in a reduction of N acquisition capacity of the whole plant in the short term (few days). Nevertheless, N2-fixing plants exhibit developmental compensatory responses to N deficit which compensate the local N limitation in the long term (> 2 wk; Jeudy et al., 2010). Indeed, both nodule formation and nodule expansion are stimulated by systemic N signaling of N deficit. These responses of nodule development to systemic N signaling have been associated with early changes in C-metabolite allocation toward N2-fixing roots at the expense of the inefficient ones.

Autoregulation of nodulation (AON) is a systemic feedback repression controlling the nodule number and the nodulation zone on the roots (see reviews of Oka-Kira & Kawaguchi, 2006 and Reid et al., 2011). Crosstalk between AON and plant N signaling has been suggested, because inhibition of nodulation by mineral N is suppressed in AON mutants, and the repressive AON can be, at least partially, released by the N demand of the whole plant in Mt (Jeudy et al., 2010). However, the response of mature nodule expansion to the plant N demand was shown to be independent of this AON regulatory circuit.

Medicago truncatula forms elongated indeterminate nodules with a persistent meristematic activity in which bacteria undergo a differentiation process from free-living reproductive bacteria to a terminally differentiated state resulting in nonreproductive, polyploid and elongated bacteroids (Mergaert et al., 2006). Histologically, indeterminate nodules of Medicago spp. are organized into five zones (Vasse et al., 1990; Timmers et al., 2000, Maunoury et al., 2010): the apical meristematic zone I, the infection zone II in which the bacteria are released from infection threads and both partners differentiate progressively, the constantly growing N2-fixing zone III, and the senescent zone IV. In older nodules, an additional zone V was described. It consists of senesced plant cells that are reinvaded by nondifferentiated bacteria, released from the remaining infection threads, which can be regarded as saprophytes.

Medicago truncatula interacts specifically with two closely related species, Sinorhizobium meliloti and S. medicae (Rome et al., 1996). Genetic diversity within these species results in variability for N2-fixation effectiveness (Mhadhbi et al., 2005; Heath & Tiffin, 2007; Rangin et al., 2008; Terpolilli et al., 2008). To succeed in nodulation in competition with other compatible rhizobia, a particular rhizobium must pass different steps, from rhizosphere colonization to root infection and nodule formation. Selection of rhizobial genotypes among compatible rhizobia present in soil has been described (Bromfield et al., 1995; Hartmann et al., 1998; Laguerre et al., 2003). It has been proposed that the plant has developed mechanisms to promote and maintain mutualistic interactions with the most beneficial partners, assuming that the host is able to discriminate among nodules based on efficiency in N2 fixation and to preferentially support the most efficient ones (see reviews of Simms & Taylor, 2002; Oono et al., 2009). Pre-infection partner choice refers to the selection by the plant of the most efficient partner before the formation of functional nodules. Post-infection partner choice refers to the ability of the plant to allocate resources preferentially to the most efficient nodules. However, there is so far only little experimental support for such mechanisms. Some studies have provided indirect evidence suggesting pre-infection partner choice (Heath & Tiffin, 2009; Sachs et al., 2010b) but other studies failed to validate this theory (Simms et al., 2006; Sachs et al., 2010a). Interestingly, the recent discovery of a plant transcription factor from the legume species Phaseolus vulgaris, NF-YC1, that influences bacterial competitiveness for nodulation possibly according to N2 fixation effectiveness (Zanetti et al., 2010) opens new perspectives to decipher the molecular basis of partner choice. Several experiments with plants co-infected by multiple partners varying in symbiotic performance provided evidence for post-infection partner choice as revealed by difference in nodule size and/or in bacterial number per nodule (Singleton & Stockinger, 1983; Simms et al., 2006; Sachs et al., 2010b). However, Heath & Tiffin (2009) did not find evidence of post-infection partner choice by Mt in a similar experiment. Suppression of N2-fixation activity by Ar:O2 treatments resulted in a rapid inhibition of development of nonfixing nodules (Singleton & van Kessel, 1987; Kiers et al., 2003, 2006; Jeudy et al., 2010; Oono et al., 2011). These results suggest a local control of nodule development by N2 fixation activity, which is consistent with the concept of post-infection partner choice. The growth inhibition of nodules placed in an Ar:O2 environment was associated with a reduction of the number of reproductive bacteria both within nodules harboring either reproductive or nonreproductive bacteroids, which has been proposed as a plant sanction against defective partners (Kiers et al., 2003, 2006; Oono et al., 2011). However, no evidence of plant sanctioning leading to differences in bacterial fitness was detected when pairs of nonfixing and fixing strains were used in split-root systems (Marco et al., 2009; Gubry-Rangin et al., 2010).

Up to now, whole plant responses to variations in N supply induced by nodulation with bacterial partners varying for N2-fixation efficiency have been poorly investigated. In this study, this question was addressed using specific experimental systems allowing the discrimination between local effects, at the site of interaction with the bacteria, and responses related to the whole plant N status and whole plant N demand. Both functioning and development of the symbiotic structures were investigated. Natural variability existing within a collection of rhizobia compatible with Mt was exploited. Interaction with inefficient bacteria and artificial suppression of fixation by Ar:O2 treatments were compared. We also addressed the question of the ability of the plant to promote nodulation with the most beneficial partner according to its increased N demand, both at the level of initial partner choice (success of nodule occupancy in co-inoculated plants) and of nodule growth. Our findings support a major role of plant N signaling in mutualistic interactions between Mt and its Sinorhizobium partners.

Materials and Methods

Bacterial strains

The bacterial strains used in this study are listed in Table 1. Bacteria were grown at 28°C for 20–24 h in yeast mannitol (YM) liquid medium or for 2–4 d on YM agar (Vincent, 1970).

Table 1.   Bacterial strains used and origin
SpeciesStrainMedicago host of origin and/or relevant characteristicsGeographic originReference
S. melilotiRCR2011M. sativaAustraliaRosenberg et al. (1981)
S. meliloti2011 TH2.3RCR2011 derivative, Nod+ Fix, fixJ::Tn5 Batut et al. (1985)
S. melilotiTII7M. truncatulaTunisiaMhadhbi et al. (2005)
S. melilotiWSM1022M. orbicularisGreeceTerpolilli et al. (2008)
S. melilotiml2M. truncatulaFranceRangin et al. (2008)
S. melilotiml6M. truncatulaFranceRangin et al. (2008)
S. melilotiml7M. truncatulaFranceRangin et al. (2008)
S. melilotiml8M. truncatulaFranceRangin et al. (2008)
S. medicaemd2M. truncatulaFranceRangin et al. (2008)
S. medicaemd4M. truncatulaFranceRangin et al. (2008)

Plant growth conditions and experimental design

The Medicago truncatula Gaertn. Jemalong A17 was used in this study. The plants were grown either under aerated hydroponic conditions in tanks (two to four replicates, five to seven plants per tank) in a climatic chamber as previously described (Ruffel et al., 2008), or in pots filled with perlite substrate (three replicates, three plants per pot) and watered weekly with an N-free nutrient solution (Moreau et al., 2008) in a glasshouse under natural light (23–28°C during the day and 18–20°C during the night). Plantlets were inoculated with bacterial cell suspensions (c. 107 per plant). In hydroaeroponic conditions, plants were first grown in a medium supplemented with 0.5 mM of KNO3. In standard tanks, after 1–2 wk of growth, the medium was changed to N-free nutrient solution and the roots were then inoculated. In pot conditions, plants were inoculated at sowing. In addition to standard tanks, split-root systems (two to four replicates, five plants per system) were used. After 1 wk of growth, the root tips were cut to promote root branching. After 1–2 wk, the root systems were separated into two parts, each side being installed in a separate compartment filled with the same N-free basal nutrient solution. Different treatments aiming to limit the N intake to one side of the root system were applied. Short-term localized N-limitation treatments were applied on nodulated roots 21 days after inoculation (DAI). To remove the N source (N2) from the treated compartment, a continuous flow of 80% argon:20% O2 was applied as previously described (Ruffel et al., 2008). For long-term localized N-treatments, roots were installed in split-root systems prior inoculation. One side was then inoculated with the non-N2-fixing strain (Fix) 2011 TH2.3, and the other side with N2-fixing strains (Fix+), either as single inoculants or using mixture of strains, both side being exposed to the same nutrient solution as given above. The nutrient solutions were renewed weekly.

Mixed inoculants and strain identification

A preliminary experiment was performed in order to estimate the ratio of bacterial cells in mixed inoculants of S. medicae md4 and S. meliloti RCR2011 to get c. 50% of nodules formed with each strain. The relationship between the ratio of bacterial cells in mixed inoculants and the resulting ratio of nodules formed by each strain was established by inoculation of seven different md4:RCR2011 ratios (from 0.1 : 99.9 to 99.9 : 0.1) to plants grown in pots. Nodules were collected 36 DAI, and rhizobial strains were isolated from crushed nodules (96 per treatment) as described previously (Depret & Laguerre, 2008). The strains md4 and RCR2011 could be reliably discriminated based on difference in colony morphology on YM agar medium. The following relationship was found (r² = 0.97), Pn md4 and Pi md4 being the proportion of md4-formed nodules and md4 cells in the inoculants, respectively:

image

Therefore, this model predicted that a 10 : 90 ratio of md4:RCR2011 cells in the inoculum should lead to an equal proportion of nodules formed with each strain.

A similar approach was used for mixtures of strains RCR2011 and Fix. The two strains were discriminated based on the streptomycin resistance (100 μg ml−1) of the Fix strain. The following relationship (R² = 0.57) was found:

image

No event of co-infection within single nodules was detected with both types of mixed inoculants.

Plant harvesting and measurements

Measurements of 15N2 fixation lasting 10 min were performed on freshly excised nodulated roots as described in Ruffel et al. (2008). This previous study reported that measurements of 15N2 fixation on excised nodulated roots were equivalent to those obtained on intact plant roots. All plant organs were collected, and nodules were counted. Dry weights were determined after oven drying at 80°C for 48 h. In some experiments, freshly collected nodules were scanned for further image analysis and used for microbiological measurements. Nodules were also conserved in Bouin’s fixative for histological analysis. Nodule size was estimated either from the mean nodule DW or nondestructively by image analysis on freshly collected nodules. A significant (R2=0.76) linear relationship between nodule DW (mg) and projected surface area (S in cm2) of nodule was obtained and is as follows: DW = 3.95 × S + 0.99. The N nutrition index (NNI) was calculated as described in Moreau et al. (2008) to quantify plant N nutrition level as the ratio between shoot N concentration and %Nc, which is the minimal N concentration allowing a maximum biomass production, estimated as %Nc = 8.1 × Shoot DW−0.10. A NNI value of 1 is considered as optimal.

Bacterial cell counting and histological methods

Single freshly collected nodules were surface disinfected and carefully crushed in sterile water. For counting of cultivable bacterial cells, nodule suspensions were serially diluted in 10-fold dilutions and plated on YM agar plates. Nodule suspensions were kept frozen in 12% v/v glycerol at − 20°C for direct cell counting by light microscopy using cell counter slides (KOVA® Glasstic® Slide) according to the manufacturer’s instruction (Hycor Biomedical GmbH, Kassel, Germany). Bacterial cells of length ≥ 2 μm were considered as elongated differentiated bacteroids according to Mergaert et al. (2006). Bacterial cells of 1–2 μm-long were considered as reproducible undifferentiated cells, which was confirmed by comparison of direct counting and counting of cultivable cells. For nodule histological analysis, semi-thin (7 μm) longitudinal sections of paraffin-embedded nodules were stained with 0.1% (w/v) toluidine blue in a 1% borax solution, observed and photographed using an Olympus Provis microscope.

Statistical analyses

Data analysis by ANOVAs and correlation was performed using XLSTAT software v2011.1.03 (Addinsoft, Paris, France). Means were classified using post hoc Least-Significant Difference (LSD) Fisher test. Log10 transformations of data were used for nodule surface area and number of cells per nodule and the logit transformation for proportions.

Results

Nitrogen acquisition by Mt varies according to the N2-fixation capacity of the symbiotic partner

The bacterial variability for N acquisition by the symbiotic association was investigated by screening a collection of nine S. meliloti and S. medicae strains from various origins which were chosen to represent the known range of variation in N2-fixation activity in symbiosis with Mt Jemalong A17 (Table 1 and references therein). This collection included the reference strain S. meliloti RCR2011 which has been reported to be poorly effective in N2-fixation with Mt Jemalong A17 (Mhadhbi et al., 2005). The efficiency of N2 fixation on the basis of nodule construction cost, expressed as SNA, was measured but considered an approximation. Although the strain-dependent variations in N accumulated in shoot g−1 nodule DW are to some extent the result of their overall symbiotic efficiency, the precise evaluation of N2 fixation efficiency, which is the ratio of benefit to cost, requires direct measurements of carbon allocation to nodule as well as respiratory loss of carbon (Oono & Denison, 2010). Significant positive linear relationships were found between SNA, N intake, plant N content, plant biomass, and nodule size (Fig. 1; Supporting Information Fig. S1 and Table S1), showing that the bacterial strain strongly affects plant N acquisition, which is a limiting factor of plant growth in these experimental conditions. Nodule number was negatively related to nodule size. The variability and ranking of strains for plant growth effect were confirmed with older plants (51 DAI) grown in standard pot conditions (Supporting Information Table S2). Plant N nutrition levels were estimated by NNI values (as defined in Moreau et al., 2008). The NNI values varied from 0.4 to 0.8 among the different symbiotic associations (Tables S1, S2), the theoretical optimal value to fulfill the plant N requirements being 1. Based on its overall higher symbiotic performance including higher SNA, the S. medicae strain md4 was selected for further comparative studies with the reference strain RCR2011.

Figure 1.

Variability of whole plant biomass and N2-fixation-specific activity of nodules according to the rhizobial partner and correlation between the two parameters (19 d after inoculation). Each point is the mean value for each strain (= 6). LSD values are 34.9 and 53.6 for plant dry weight (DW) and specific nodule activity, respectively. The statistical comparison of means is given as supplementary data in Table S1.

Effect of the bacterial partner on the plant response to localized N limitation

Responses of plants inoculated with md4 or RCR2011 to localized N limitation were characterized using plants grown in split-root systems. Adaptation to N deficit was characterized by comparing N acquisition and developmental responses of N-limited and control plants.

Short-term N limitation treatments (4 d) were applied 21 DAI on nodulated half-root systems by replacing air with a mixture of 80% Ar and 20% O2 for 4 d (Fig. 2a). The response to plant N deficit of the nodulated half-root side that remained supplied by air (Fix+) was investigated. The whole plant N intake by the N-limited plants was 54% and 41% lower than in the control plants with md4 and RCR2011, respectively (Fig. 2b). Consequently, shoot N content was also reduced in N-limited plants by 42% and 22% with md4 and RCR2011, respectively (Fig. S2). As expected, nodule SNA in control Fix+ half-roots nodulated by md4 was higher than in RCR2011-nodulated half-roots. These activities remained unchanged in response to the N deficit treatment (Fig. 2c) indicating the absence of compensatory increase of SNA under such conditions. However, nodule expansion was stimulated with md4 in response to N deficit (23% higher than the control plants; Fig. 2d).

Figure 2.

Effect of the bacterial strain on responses of N2-fixing plant to short-term (4 d) local N limitation induced by suppression of N2 from the root atmosphere by Ar:O2 treatment of nodulated roots on one side of split-roots systems shown in (a). The N intake by the whole plant (b), specific nodule activity (c), and mean nodule size (d) of nontreated half-root systems were compared between the control (dark grey bars) and N-limited plants (light grey bars). Half-root systems exposed to the same local environment during the treatment (central compartment) were used for comparison. Values are the means of 10 replicates. Identical letters show that the mean values are not significantly different by LSD (P ≤ 0.05).

Long-term N limitation treatments (30 d) were applied by inoculating one side of split-root systems (Fix side) with a Fix (fixJ) mutant of RCR2011, the other side (Fix+) being inoculated with md4 or RCR2011 (Fig. 3a). In Fix roots, the fixJ mutation did not alter the ability of the bacteria to form nodules, the nodule number per half-root being equivalent to that in control half-roots nodulated by the parental strain (data not shown). The Fix nodules were white and smaller than those formed with RCR2011 in control plants. The response to plant N deficit of the nodulated half-root side that remained supplied by air (Fix+) was investigated. For each Fix+ strains, the N-limited and the control plants displayed similar biomasses (Fig. 3b) and N contents despite the absence of efficient N2-fixation in Fix half-roots (Table S3). The N-limited plants were able to fully compensate for the local N limitation induced by ineffective nodulation in half-root systems (Fix sides) by increasing the N acquisition capacity of the half-root systems of the Fix+ sides (62–64% higher than in control roots; Table S3). This long-term compensatory response resulted from a strong increase of nodule biomass in Fix+ roots explained by an increase of both nodule number and mean nodule size (Fig. 3) rather than from adjustment of nodule SNA. Biomass of Fix+ roots was also increased (Table S3). Both symbiotic associations displayed qualitatively similar responses, but these responses were more efficient with md4 than with RCR2011. Although the increase in nodule number in Fix+ roots was comparable for all strains (33–37% higher than in control roots), the stimulation of nodule expansion in Fix+ roots was stronger with md4 than with RCR2011 (70% and 30%, respectively). The Fix+ nodules of N limited plants were distributed into two classes of size. The smallest nodules displayed a size similar to that of nodules of control plants (Fig. 4). They were probably young nodules formed de novo in response to N limitation. The analysis of nodule structure (zonation and bacterial infection) showed that the biggest nodules, probably representing the expanded nodules, displayed a ‘peanut’ shape (Fig. 4b,d). This shape suggests two waves of nodule development as if, at some point, the perception of N deficiency caused an acceleration of nodule development. This response was much stronger in nodules formed with md4 (Fig. 4d) than in those formed with RCR2011 (Fig. 4b). The expansion of md4 nodules corresponded to a significant increase of the N2-fixing zone III surface and of the number of symbiotic plant cells, while we did not record such significant responses in elongated nodules formed with RCR2011.

Figure 3.

Effect of the Fix+ bacterial strain applied as single inoculant on responses of N2-fixing plants to long-term (30 d) local N limitation induced by nodulation of half-root systems with a Fix strain in split-roots systems shown in (a). Absence of N2-fixation activity was confirmed for the Fix roots. (b) Entire plant biomass. Developmental responses were compared between the control (dark grey bars) and N-limited plants (light grey bars) on the half-root systems exposed to the same local environment during the treatment (central compartment): (c) nodule biomass, (d) number of nodules, (e) mean nodule size. Values are the means of 10 replicates. Identical letters show that the mean values are not significantly different by LSD (P ≤ 0.05).

Figure 4.

Nodule structure (30 d after inoculation) in plants grown in split-root systems (see Fig. 3a) and inoculated either with RCR2011 or md4 on half-root systems. Semi-thin longitudinal nodule sections were stained with toluidine blue and observed by light microscopy. (a) RCR2011 in control plants, (b) RCR2011 in N-limited plants, (c) md4 in control plants, (d) md4 in N-limited plants. 1, meristem; 2, infection and differentiation zone II; 3, nitrogen fixation zone III. Bars, 200 μm.

Effect of the N status of the plant on the symbiotic partner preference

The influence of the plant N status on initial partner choice, as revealed by successful nodulation, was investigated by inoculation of plants grown in hydro-aeroponic conditions with a mixture of md4 and RCR2011 cells. In a preliminary experiment where the two strains were co-inoculated in different cell ratios, md4 was found to more competitive for nodulation than RCR2011 in absence of mineral N fertilization (see Material and Methods). A 10 : 90 ratio of md4:RCR2011 cells in the inoculum was chosen to obtain an equal proportion of nodules formed by each bacterial strain. The effect of 0.2 or 2 mM KNO3 supplies to the plants on the md4:RCR2011 nodule ratio was investigated. The contrasted N regimes resulted in significant differences in plant biomass, plant N content, and NNI (Fig. 5). The plant N status also had a strong effect on nodule formation as the number of nodules was four-fold higher in the 0.2 mM KNO3 treatment (Fig. 5c). Nevertheless, the identification of the nodulating bacteria and the quantification of the proportion of nodules formed with each strain in these assays showed that the N regimes tested here did not significantly affect the initial partner choice. Indeed, 63% and 53% of nodules were formed by md4 for 2 and 0.2 mM KNO3 treatments, respectively (differences no statistically significant at the 0.05 probability level).

Figure 5.

Effect of the plant N status of N2-fixing plants supplied with 2 or 0.2 mM KNO3 and inoculated with a mixture of strains md4 and RCR2011 on plant biomass (a), shoot N content (b), and nodulation (c). Mean values of N nutrient index (NNI) were of 0.89 and 0.73 for plants fed with 2 mM and 0.2 mM KNO3, respectively. The measurements were carried out 20 d after inoculation. Values are the means of 12 replicates. Identical letters show that the mean values are not significantly different by LSD (P ≤ 0.05).

The plant N regime had an important effect on nodule expansion as nodule size was greater in those plants supplied with 0.2 mM KNO3 than in those supplied with 2 mM KNO3 (Fig. 6a). This effect was stronger for md4 nodules than for RCR2011 nodules (increase in size of 162% and 58%, respectively). Interestingly, nodule expansion resulted in a highly significant stimulation of the multiplication and elongation of differentiated bacteroids in md4 nodules, but not in a significant increase in numbers of cultivable bacteria per nodule (Fig. 6b–d; Table 2).

Figure 6.

Effect of the bacterial strain on nodule development according to the N status of N2-fixing plants supplied with 2 mM KNO3 (dark grey bars) or 0.2 mM KNO3 (light grey bars) and inoculated with a mixture of strains md4 and RCR2011. The measurements were carried out 20 d after inoculation: (a) mean nodule size, (b) mean number of cultivable bacteria per nodule, (c) mean number of differentiated bacteroids (length of cells > 2 μm) per nodule, (d) mean length of differentiated bacteroids. Values are the means of individual measurements of 40–80 nodules, with measurements on 60–100 individual bacteroids per nodule. Identical letters show that the mean values are not significantly different by LSD (P ≤ 0.05).

Table 2.   Matrix of correlationsa among nodule sizes, numbers of cultivable bacteria and differentiated bacteroids, and bacteroid length 20 d after inoculation in N2-fixing plants co-infected with md4 and RCR2011 and supplied with 2 or 0.2 mM KNO3
Variable2 mM KNO30.2 mM KNO3
(1)(2)(3)(4)(1)(2)(3)(4)
  1. aValues above and below diagonal are Pearson correlation coefficients obtained for strains md4 and RCR2011, respectively. Mean values over strain/KNO3 treatments (= 21–22 for md4; = 13–14 for RCR2011).

  2. bLog10 transformed values. Correlations are significant at *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001 respectively.

Nodule sizeb (1)1− 0.160.300.47*10.43*0.70***0.61**
No of cultivable cells per noduleb (2)0.77***1− 0.39− 0.45*0.2710.0− 0.08
No of bacteroids per noduleb (3)0.270.5010.81***0.65*0.2010.70***
Bacteroid length (4)0.54*0.410.2610.28− 0.300.74**1

Additional split-root experiments were designed to discriminate between the effects of the local and systemic N signalings. The system used was similar to the one described above for investigation of long-term N-limitation except that a mixture of md4:RCR2011 was used instead of single inoculants (Fig. S3). The absence of a systemic effect of the N status of the plant on initial partner choice was confirmed as Fix+ roots of N-limited and control plants displayed identical md4:RCR2011 nodule ratios (Table S4). Again, the stimulation of nodule expansion in response to systemic N signaling was higher with md4 than with RCR2011, and the expansion of md4 nodules was associated with a significant increase of the mean length of bacteroids and of the number of bacteroids per nodule. The number of cultivable bacteria per nodule did not vary among treatments and among bacterial strains (Table S4). In half-roots nodulated by the Fix strain, the mean number of undifferentiated bacteria per nodule (5 × 105 cells) and bacteroid length (5 μm) were equivalent to those obtained in Fix+ nodules of control plants. However, the mean number of bacteroids per Fix nodule was severely affected by the fixJ mutation (mean values of 5 × 104 and 2 × 106 cells for the Fix nodules and the Fix+ nodules of control plants, respectively).

All together these data suggest that the plant does not favor the best partner at early stages of nodulation in response to an increase in N demand. At later stages, systemic signaling of N deficit might stimulate the development of nodules according to the bacterial strain in symbiotic organs. The preferential growth stimulation of md4 nodules in co-inoculated plants is not likely to be at the expense of RCR2011 nodules because the growth of RCR2011 nodules in these conditions was not repressed in response to N deficit (Table S4).

Role of spatial heterogeneity of N provision on the root system in the plant response to N deficit

In the split-root experiments described in this study, variations in the N status of the plant were obtained by locally suppressing N2-fixation in nodulated half-root systems (Ar:O2 treatment or nodulation by a Fix rhizobial strain). It remained unclear whether the global plant N deficit or the imbalance in N provision between spatially separated root systems is sufficient by itself to trigger the plant compensatory response. In order to address this issue, the split-root plants in which the Fix nodulated roots were spatially separated from the Fix+ (RCR2011) nodulated roots were compared with plants inoculated with a mixture of RCR2011 and the Fix strain on the whole root system (Fig. 7a,b). The data from the split-root experiment are those given above (Fig. 3; Table S3). Control plants inoculated with single inoculants were also included in the comparison. In co-infected plants, the nodules formed by each strain were homogeneously distributed on the whole root system (see Fig. S4), 20% of the nodules being formed by the Fix strain. A weak expansion (+15%) of RCR2011 nodules was observed in these plants by comparison with the RCR2011 control plants (Fig. 7d). However, the whole plant biomass in these plants was reduced by 32% by comparison with the RCR2011 control plants (Fig. 7c) showing that the response to N deficit was not as effective than in split-root plants. Although 50% of nodules were Fix, the split-root plants were still able to fully compensate the N deficit by specifically stimulating expansion of nodules on the Fix+ half-root systems. This result strongly suggests that the effectiveness of the compensatory response to N deficit of split-root plants was dependent on the imbalance of root intake within the root system due to spatial separation of the Fix nodules. The suppression of such a discrete distribution of RCR2011 and Fix nodules drastically reduced plant ability to compensate efficiently the N deficit, probably because the systemic response to the N deficit was much less effective when diluted on the whole root system.

Figure 7.

Effect of homogenous co-infection by RCR2011 and the Fix strain of the whole root system on plant and nodule development. The measurements were carried out 26 d after inoculation: (a) Split-roots systems with spatially separated RCR2011 and Fix nodulated roots used for comparison (56% of RCR2011 nodules at the whole plant level). (b) Experimental design used for plants homogenously co-infected on the whole root system (80% of RCR2011 nodules) including control plants mono-infected with RCR2011 and the Fix strain. A 30 : 70 ratio of RCR2011:Fix cells was used as inoculants. (c) Entire plant biomass (mean values of 16 replicates), (d) mean nodule size based on individual measurements of 220–270 RCR2011 nodules and 70–120 Fix nodules. RCR2011 and Fix nodules from co-infected plants (light grey bars) were compared with nodules of their respective control (dark grey bars) (mono-infected plants). Identical letters indicate that the mean values are not significantly different by LSD (P ≤ 0.05).

The plants that were co-inoculated with the mixture of RCR2011 and the Fix strain were also compared with the Fix control plants (Fig. 7b). As expected, the absence of N2 fixation in the Fix plants resulted in a severe N limitation and a reduced plant growth (Fig. 7c). The size of the Fix nodules was smaller in these plants than in the co-infected plants (Fig. 7d). The presence of Fix+ nodules on the root system of co-infected plants improved plant N nutrition, which probably made possible the expansion of the Fix nodules. This result indicates that the Fix nodules benefited locally from more resource allocation in plants that were supplied in N by the RCR2011 nodules. Similar results were obtained using mixtures of RCR2011 and md4, the growth of RCR2011 nodules benefiting from the occurrence of md4 nodules on the root system (Table S5).

Discussion

In this study, we have characterized the N nutrition of Mt Jemalong A17 in symbiosis with various Sinorhizobium strains. The plant growth and the N acquisition rate of these associations were correlated to SNA, showing that, even with the most effective partner, N2 fixation remains the limiting factor of plant growth in our conditions. Symbiotic N nutrition was always suboptimal compared with plants supplied with mineral N, consistent with previous reports on Mt Jemalong A17 (Ruffel et al., 2008; Moreau et al., 2008; Sulieman & Schultze, 2010). Whether the suboptimal N acquisition is a general phenomenon within the Mt-Sinorhizobium associations or is specifically related to the Jemalong A17 line, remains to be investigated.

We show that the plant responses to partial nodulation with a Fix strain involve whole plant N signaling mechanism(s). N deficit resulting from the Fix nodules in a half-root system triggered a compensatory response that stimulates the N acquisition by the other half-root system when nodulated with a Fix+ bacterium. This response was able in the long term to fully compensate the N deficit of the whole plant. The increased N acquisition was related to the stimulation of nodule development by systemic N signaling. Both nodule initiation and nodule expansion in the Fix+ side were instrumental to this response with both strains RCR2011 and md4. However, there was no compensatory increase in SNA in response to N deficit even with the most effective strain (Fig. 2c; Table S3). This result is consistent with the assumption that SNA would always be at its maximum in both N-limited and control plants as previously proposed (Ruffel et al., 2008; Jeudy et al., 2010). Taken together, the present data indicate that the model of regulation of symbiosis by the plant N demand initially based on the effect of artificial Ar:O2 treatments on the Mt-RCR2011 association (Jeudy et al., 2010) also works for a more efficient symbiotic association. Our study shows that the plant response to the variability of the bacterial partner can be explained by this model.

Our data suggest that the plant probably addresses systemic N signal(s) in response to N limitation at the scale of individual root bundles that are globally perceived as efficient or inefficient rather than at the scale of individual nodules. The efficiency of the adaptative response of the whole plant to partial suppression of N2 fixation was related to the spatial separation of the N limitation within the root system. When Fix+ and Fix nodules were homogeneously distributed on the roots of co-infected plants, the plant responded by increasing the size of the Fix+ nodules. However, this response was not great enough to fully compensate the N deficit caused by the Fix nodules. These results are consistent with early work by Singleton & Stockinger (1983) reporting a compensatory expansion of Fix+ nodules in an equivalent experiment with soybean, but also a negative relationship between N acquisition and the proportion of Fix nodules. Our study suggests that in absence of imbalanced N provision between root bundles, the systemic response to the N deficit is diluted on the whole root system, which limits the efficiency of the response. Earlier work by Jeudy et al. (2010) has shown that the compensation of local suppression of N2 fixation by Ar:O2 treatments was associated with a preferential allocation of carbon metabolites to half-root systems that remained efficient for N2 fixation at the expense of the inefficient ones. Therefore, a plausible explanation for our results could be that the plant can neither discriminate between effective and ineffective structures nor allocate its limited resources only to the effective ones in absence of spatial separation of these structures. Moreover, this assumption is also consistent with our results, suggesting that the less effective partner benefits indirectly from the most effective ones of more resources. This phenomenon would contribute to the persistence of ineffective rhizobia in soils.

Although plants associated with either md4 or RCR2011 shared the same general response to local N limitation, this response was more efficient with md4 than with RCR2011. This result was partly expected because the higher SNA in md4 nodules would lead by itself to a higher level of N acquisition even if nodule biomass had similarly increased with both strains. However, the stronger expansion of md4 nodules contributed additionally to increase the N acquisition level. Histological analyses of md4 nodules indicated that the number of symbiotic plant cells was increased in response to systemic N signaling, suggesting that nodule meristematic activity and differentiation of symbiotic plant cells are stimulated in mature nodules. No evidence for such stimulation was seen with RCR2011. These data are consistent with previous work suggesting an influence of the bacterial partner on the nodule meristematic activity (Sagan & Gresshoff, 1996; Tirichine et al., 2000; Laguerre et al., 2007). The preferential response of md4 nodule expansion was also correlated to a stimulation of bacteroid differentiation and elongation. In contrast to nodule expansion, stimulation of nodule formation in response to systemic N signaling remained quantitatively equivalent with both strains, indicating that this process was probably not influenced by the bacterial partner in these conditions. This result is in agreement with a genetic study indicating that the responses of nodule formation and nodule expansion to systemic N signaling are probably mediated by distinct pathways (Jeudy et al., 2010).

The mechanisms underlying the strain-dependent responses of mature nodule development to systemic N signaling deserve further investigation. The fact that strain- dependent differences were observed on co-infected roots where md4 and RCR2011 nodules or Fix and Fix+ nodules were in vicinity and were likely to receive the same phloem allocation indicates that nodule specific factors have an important role in determining locally the response of nodule development to systemic signaling. The relation at the scale of the individual nodule between the rates of N2 fixation and of stimulation of nodule growth in response to systemic N signaling suggests a possible role of nodule SNA in the control of nodule expansion. However, additional traits may be involved and it remains to be elucidated whether they are determined by the host plant or/and by the bacterial partner within the symbiotic organ.

In addition to systemic signaling resulting in stimulation of development of effective symbiotic structures in response to local N limitation, N2 fixation may exert a local control on nodule development. Several studies on diverse legume species have supplied evidence for a rapid inhibitory effect of Ar:O2 treatments on nodule growth at the site of application (Singleton & van Kessel, 1987; Jeudy et al., 2010; Oono et al., 2011), including at the scale of individual nodules (Kiers et al., 2003). It has been proposed that the plant develops mechanisms to locally restrict the development of nodules formed with ineffective bacteria (Kiers et al., 2003; Simms & Taylor, 2002; Oono et al., 2009, 2011). In the long term, Ar:O2 treatments resulted in a decrease of the number of cultivable bacteria in nodules, which could be associated with early nodule senescence, in legumes forming nodules harboring either reproductive or nonreproductive bacteroids (Kiers et al., 2003, 2006; Oono et al., 2011). This result was interpreted as a host sanction towards the less beneficial partners. However, in the present study, although the absence of N2 fixation resulted in restriction of nodule development, we did not find evidence of sanction towards the Fix nodules based on counting of reproductive undifferentiated bacteria. This observation is in agreement with previous reports on analogous experiments (Marco et al., 2009; Gubry-Rangin et al., 2010). Also, we did not observe difference in bacterial fitness between RCR2011 and md4 nodules. Nevertheless, the absence of sanction in our experimental conditions may be related to the developmental stage of the 4-wk-old nodules analysed which did not yet show symptoms of senescence. Indeed, it has been reported that the multiplication of undifferentiated rhizobial cells in Medicago nodules was affected by the Fix mutation only in senescent plant cells reinvaded by saprophytic undifferentiated bacteria in nodule zone V (Timmers et al., 2000; Maunoury et al., 2010). An impact of sanction at the latest stage of nodule growth, when reproducible bacteria will be released in the soil, would be consistent with a function that tends to limit the multiplication and the dispersion of ineffective rhizobia.

Our study does not provide evidence that the plant N demand influences the initial partner choice, at least between md4 and RCR2011. Our results clearly show that the plant N status has no effect on their competitive success in nodulation suggesting that, whatever the mechanisms behind competitiveness for nodulation, they operates in a same way under different N status. Consistently, systemic N signaling also stimulated nodule formation similarly for both strains (Fig. 3). This result is in agreement with earlier co-inoculation studies with Fix+ and Fix strains showing that N2-fixation per se does not influence the success of nodule formation (e.g. Johnston & Beringer, 1976; Amarger, 1981).

In conclusion, both local and systemic signaling mechanisms are likely to contribute concomitantly to the whole plant adaptation to fluctuation of its root N supply due to variations of effectiveness of its bacterial partners. We propose a general model where both mechanisms modulate the interactions between the host plant and its symbiotic partners. The whole plant N demand triggers systemic controls regulating nodule development and functioning. When the plant N demand is fully satisfied (sufficient addition of mineral N), an ‘N satiety’ systemic repression of N2 fixation activity operates (Ruffel et al., 2008). However, when the plants rely exclusively on N2 fixation, this repression is not active because under such conditions the plant N nutrition is suboptimal. In these plants, a more severe whole plant N deficit triggers the ‘N deficit’ systemic stimulation of nodule formation and expansion. The integration by the plant of effectiveness of root N acquisition in this process is likely to be at the scale of the root bundle rather than at the scale of the individual nodule. Together, ‘N satiety’ and ‘N deficit’ behave as systemic feedback mechanisms where whole plant N assimilates modulate N acquisition capacity and development of the symbiotic structures. These regulatory loops are tightly integrated with the AON repression mechanism which is involved in the repression of nodulation by ‘N satiety’ signaling, and partially released under ‘N deficit’ (Jeudy et al., 2010). In addition to systemic signaling, there are also local controls operating at the individual nodule level. First, the expansion of effective symbiotic structures at mature stages is preferentially stimulated by the ‘N deficit’ systemic signaling. Second, low N2-fixation activities result in a repression of growth of the symbiotic structures and likely in a decrease of bacterial fitness when bacteria are released in the environment. This nodule-specific mechanism might include the ‘plant sanction’ initially described by Kiers et al. (2003). Whether growth stimulation of effective nodules and growth repression of ineffective ones are particular aspects of a general mechanism coupling nodule growth and nodule activity at the local level, or belong instead to specific mechanisms deserve further studies.

Acknowledgements

We thank Cécile Gubry-Rangin for useful advice on rhizobial strain selection. We are grateful to Naïma Rezkallah for help with microbiological work, and to Guy Ruiz and Marc Boursot for technical assistance in the plant experiments. We also thank Eric Giraud and Michel Lebrun for helpful comments on the manuscript. This work was funded by Agropolis Foundation and INRA.

Ancillary